Project: Extraction and transfer of knowledge contained in trained Artificial Neural Networks: development of a methodology for the classification of tropical vegetation using digital spaceborne data.
- project duration
- 01-JAN-02 – 31-OCT-04
- The overall aim of this project is to investigate how techniques developed in the information theory can be used to overcome the disadvantages of neural networks. Discriminant Based Transfer, Rule Extraction and Knowledge Based Artificial Neural Networks will be tested for their suitability, within the context of tropical vegetation mapping.